✒✒Microsoft unveils trOCR, the neural evolution of OCR via Transformers. Source code available!✒✒

👉Optical Character Recognition (OCR) is the electronic...

💹💹NVIDIA #AI unveils the most advanced framework to estimate physically correct human motions💹💹

👉Human motion synthesis is an important problem with...

500 million+ members | Manage your professional identity. Build and engage with your professional network. Access knowledge, insights and opportunities.

🧊🧊Facebook unveils a Transformer-based #AI able to 𝗽𝗿𝗲𝗱𝗶𝗰𝘁 𝘁𝗵𝗲 𝗳𝘂𝘁𝘂𝗿𝗲🧊🧊

👉Predicting future human actions is an important task for...

🍧🍧NVIDIA unveils CLIPort, a super #AI agent mixing CLIP (OpenAI) and TransporterNets (Google)🍧🍧

👉Recent works in manipulation have shown that end-to-end...

🥣🥣University of Oxford releases 𝐏𝐀𝐒𝐒 dataset: 1.4M+ images (free from humans) for self-supervised #deeplearning🥣🥣

👉PASS is a large-scale image...

🙋‍♂️🏎Lyft unveils the world's most advanced simulating driving experiences for #selfdrivingvehicles🙋‍♂️🏎

👉Self-Driving Vehicles (SDVs) have the potential...

🏄‍♂️🏄‍♂️ETH Zürich unveils FBCNN to restore JPEG images surfing between artifacts removal and details preservation 🏄‍♂️🏄‍♂️

👉Training a single deep...

💥💥Baidu, Inc. unveils PP-LCNet (lightweight CPU-CNN😦) and greatly outperforms the most of the SOTA models. BOOM! 💥💥

👉A lightweight CPU network based... 18 comments on LinkedIn

😵😵Microsoft open-sourcing a huge synthetic-faces dataset. Website announced a few mins ago!😵😵

👉Is it possible to perform SOTA face-related #computervision... 32 comments on LinkedIn

♻♻Koreans researchers and SAMSUNG SDS unveil ILVR, a novel generative process in DDPM for HQ images♻♻

👉Generative models, such as GAN, normalizing flows...

🎥💥Shanghai, 1929: #AI Restoration by Glamourdaze. Magic 🎥💥

🔈🔉🔊 SOUNDS ON 🔈🔉🔊

Highlights:
✅Cleaned noise artifacts
✅Increased framerate from 24 to…

☑☑ PySlowFast library by Facebook AI just added the neural Transformers support ☑☑

👉PySlowFast is an open source video understanding codebase from FAIR...

💥💥From #ICLR22, StyleNeRF: a novel #AI to synthesize HD images preserving 3D consistency💥💥

👉Photo-realistic free-view image synthesis of real-world... 16 comments on LinkedIn

🥝🥝UOAIS: a novel #AI to segment unseen objects via HOM scheme🥝🥝

👉Segmenting objects is an essential skill for robotic manipulations in an unstructured...

♻♻HARP: autoregressive latent video prediction #AI model for predicting High-Res future frames♻♻

👉Video prediction can enable agents to learn useful ...

🍧🍧Microsoft unveils the source code of Mesh Graphormer, a new Transformer-based #AI🍧🍧

👉3D human pose and mesh reconstruction from a single image ...

😵😵Huawei unveils HeadGAN, head reenactment, disentangling identity from expression😵😵

👉Recent attempts to solve the problem of head reenactment using... 15 comments on LinkedIn

🧁🧁CMU + NVIDIA open sourcing a novel self-supervised autoencoder for video generation🧁🧁

👉The art of seeing is in our ability to disentangle the reality...

🛸🛸 University of Zurich + Intel AI open-sourcing a framework for high-speed autonomous flight through challenging environments 🛸🛸

👉Quadrotors are... 12 comments on LinkedIn